LLM Reference

Phi-3 Medium 128K

Released
2024-05-21
Last refreshed
2026-05-16
Status
Researched 189d ago
Open sourceCommercial use: permittedCodingLong contextClassification

Phi-3 Medium 128K is worth evaluating for coding, long context, and classification when its provider route and context window match the workload.

Use it for

  • Teams evaluating coding, long context, and classification
  • Workloads that can use a 128k context window
  • Buyers comparing 2 tracked provider routes

Do not use it for

  • Vision or document-understanding workloads
  • Strict JSON or tool-calling flows
Specifications
Family
Phi-3
Released
2024-05-21
Context
128k
Parameters
14B
Architecture
Decoder Only
Knowledge cutoff
2023-10
Specialization
general
Openness
Open source
License
MITOSI-approvedCommercial use: permitted
Weights
Unknown
Code
Unknown
Training
Fine-tuned
Created by

Advancing the state-of-the-art in AI and computing.

Redmond, Washington, United States
Founded 1991
Website
Pricing
Output / 1M
$1.50
Input / 1M
$0.500

Cheapest of 2 routes · Microsoft Foundry

About

The Phi-3 Medium 128K is an open-source, 14-billion parameter language model by Microsoft, designed for efficient operation in resource-limited environments. Noted for its state-of-the-art performance on reasoning tasks, it excels in language understanding, code generation, and logical reasoning while offering a long context window of up to 128,000 tokens, making it ideal for applications like summarizing lengthy documents. Its dense decoder-only Transformer architecture has been refined with supervised fine-tuning and preference optimization to enhance instruction-following capabilities. Additionally, Phi-3 Medium 128K is optimized for diverse hardware platforms, ensuring broad accessibility and performance 12.

Phi-3 Medium 128K is an open-source model in the Phi-3 family. The structured metadata tracks a 128k-token context window. This page tracks provider routes through Microsoft Foundry and NVIDIA NIM, with the cheapest tracked route listed at $0.5 input and $1.5 output per 1M tokens. Headline tracked benchmarks include HumanEval 52.2, Massive Multitask Language Understanding 75.3, and MMLU PRO 51.9.

Top use-case fit: coding, agents, and build tasks

Coding

Q/$ C

1 relevant benchmark in the decision map.

Long context

Included by capability and metadata signals in the decision map.

Classification

Q/$ D

2 relevant benchmarks in the decision map.

Provider price ladder

Compare all 2

Compare API pricing across 2 providers for input and output tokens, batch, and cached reads when available.

ProviderInput / 1MOutput / 1MRoute
Microsoft Foundry$0.500$1.50
ServerlessProvisioned
NVIDIA NIM--
ProvisionedPartial

Available via routers & gateways(6)

Capabilities

No model capability flags are currently sourced.

Benchmark peer barsfor Coding

Benchmark scores(3)

Scores are benchmark-specific and are direction-aware: the same numeric gap can mean very different outcomes across suites. Use the leaderboard context and this model's provider route to decide whether the winning margin is meaningful for your workload.
BenchmarkScoreVersionSource
HumanEval52.2pass@1https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard
Massive Multitask Language Understanding75.35-shothttps://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard
MMLU PRO51.9https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro

Migration checks

No linked migration route is available for this model yet.

Compare Phi-3 Medium 128K with other models

Frequently asked questions

What is the context window of Phi-3 Medium 128K?

Phi-3 Medium 128K has a context window of 128k tokens.

How much does Phi-3 Medium 128K cost?

Phi-3 Medium 128K is available at $0.5/1M input tokens through Microsoft Foundry.

When was Phi-3 Medium 128K released?

Phi-3 Medium 128K was released on 2024-05-21.

Which providers offer Phi-3 Medium 128K?

Phi-3 Medium 128K is available from 2 providers: Microsoft Foundry, NVIDIA NIM.

What benchmarks has Phi-3 Medium 128K been tested on?

Phi-3 Medium 128K has been evaluated on 3 benchmarks, including HumanEval, Massive Multitask Language Understanding, MMLU PRO.